A renaissance of neural networks in drug discovery

Baskin, Igor I.; Winkler, David A.; Tetko, Igor V.

doi:10.1080/17460441.2016.1201262

Cited by 208 publications

(159 citation statements)

References 109 publications

Supporting

Mentioning

149

Contrasting

Unclassified

Order By: Relevance

“…Previously deep learning has been used mostly for unsupervised learning and noisy data 28–34 . Limited efforts to use deep learning for pharmaceutical applications suggest a need for further exploration to access its utility for cheminformatics compared with other methods 35 . Deep Learning has been relatively widely used for bioinformatics 36 and computational biology 37 .…”

Section: Introductionmentioning

confidence: 99%

Comparison of Deep Learning With Multiple Machine Learning Methods and Metrics Using Diverse Drug Discovery Data Sets

Korotcov¹,

Tkachenko²,

et al. 2017

View full text Add to dashboard Cite

Machine learning methods have been applied to many datasets in pharmaceutical research for several decades. The relative ease and availability of fingerprint type molecular descriptors paired with Bayesian methods resulted in the widespread use of this approach for a diverse array of endpoints relevant to drug discovery. Deep learning is the latest machine learning algorithm attracting attention for many of pharmaceutical applications from docking to virtual screening. Deep learning is based on an artificial neural network with multiple hidden layers and has found considerable traction for many artificial intelligence applications. We have previously suggested the need for a comparison of different machine learning methods with deep learning across an array of varying datasets that is applicable to pharmaceutical research. Endpoints relevant to pharmaceutical research include absorption, distribution, metabolism, excretion and toxicity (ADME/Tox) properties, as well as activity against pathogens and drug discovery datasets. In this study, we have used datasets for solubility, probe-likeness, hERG, KCNQ1, bubonic plague, Chagas, tuberculosis and malaria to compare different machine learning methods using FCFP6 fingerprints. These datasets represent whole cell screens, individual proteins, physicochemical properties as well as a dataset with a complex endpoint. Our aim was to assess whether deep learning offered any improvement in testing when assessed using an array of metrics including AUC, F1 score, Cohen’s kappa, Matthews correlation coefficient and others. Based on ranked normalized scores for the metrics or datasets Deep Neural Networks (DNN) ranked higher than SVM, which in turn was ranked higher than all the other machine learning methods. Visualizing these properties for training and test sets using radar type plots indicates when models are inferior or perhaps over trained. These results also suggest the need for assessing deep learning further using multiple metrics with much larger scale comparisons, prospective testing as well as assessment of different fingerprints and DNN architectures beyond those used.

show abstract

Section: Introductionmentioning

confidence: 99%

Comparison of Deep Learning With Multiple Machine Learning Methods and Metrics Using Diverse Drug Discovery Data Sets

Korotcov¹,

Tkachenko²,

et al. 2017

View full text Add to dashboard Cite

show abstract

“…25 Neural network embedded chemical fingerprints have shown promise, but share limitations common to deep learning: Models are computationally expensive and difficult to train, and hyperparameters like the learning rate, smoothing parameters, and model architecture must be tuned for each application. 20, 44, 45 Shallow learning offers a simpler and more robust alternative, but with limitations. Restricting network depth makes training easier and more efficient, but limits the expressiveness of the range of nonlinear representations that can be learned.…”

Section: Discussionmentioning

confidence: 99%

Shallow Representation Learning via Kernel PCA Improves QSAR Modelability

Rensi

Altman

2017

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Linear models offer a robust, flexible, and computationally efficient set of tools for modeling quantitative structure activity relationships (QSAR), but have been eclipsed in performance by non-linear methods. Support vector machines (SVMs) and neural networks are currently among the most popular and accurate QSAR methods because they learn new representations of the data that greatly improve modelability. In this work we use shallow representation learning to improve the accuracy of L1 regularized logistic regression (LASSO) and meet the performance of Tanimoto SVM. We embedded chemical fingerprints in Euclidean space using Tanimoto (aka Jaccard) similarity kernel principal components analysis (KPCA), and compared the effects on LASSO and SVM model performance for predicting the binding activities of chemical compounds against 102 virtual screening targets. We observed similar performance and patterns of improvement for LASSO and SVM. We also empirically measured model training and cross validation times to show that KPCA used in concert with LASSO classification is significantly faster than linear SVM over a wide range of training set sizes. Our work shows that powerful linear QSAR methods can match nonlinear methods, and demonstrates a modular approach to non-linear classification that greatly enhances QSAR model prototyping facility, flexibility, and transferability.

show abstract

“…In this case, neurons are arranged in layers, internal layers are called hidden layers. ANN performances in QSPR development has been widely demonstrated …”

Section: Basic Elements For Qspr Modelingmentioning

confidence: 99%

Inverse‐QSPR for de novo Design: A Review

Gantzer

Creton

Nieto‐Draghi

2019

Molecular Informatics

View full text Add to dashboard Cite

The use of computer tools to solve chemistry‐related problems has given rise to a large and increasing number of publications these last decades. This new field of science is now well recognized and labelled Chemoinformatics. Among all chemoinformatics techniques, the use of statistical based approaches for property predictions has been the subject of numerous research reflecting both new developments and many cases of applications. The so obtained predictive models relating a property to molecular features – descriptors – are gathered under the acronym QSPR, for Quantitative Structure Property Relationships. Apart from the obvious use of such models to predict property values for new compounds, their use to virtually synthesize new molecules – de novo design – is currently a high‐interest subject. Inverse‐QSPR (i‐QSPR) methods have hence been developed to accelerate the discovery of new materials that meet a set of specifications. In the proposed manuscript, we review existing i‐QSPR methodologies published in the open literature in a way to highlight developments, applications, improvements and limitations of each.

show abstract

A renaissance of neural networks in drug discovery

Cited by 208 publications

References 109 publications

Comparison of Deep Learning With Multiple Machine Learning Methods and Metrics Using Diverse Drug Discovery Data Sets

Comparison of Deep Learning With Multiple Machine Learning Methods and Metrics Using Diverse Drug Discovery Data Sets

Shallow Representation Learning via Kernel PCA Improves QSAR Modelability

Inverse‐QSPR for de novo Design: A Review

Contact Info

Product

Resources

About